Paper # 7CO-63 Topic: Coal and Biomass Combustion and Gasification 8 th U. S. National Combustion Meeting Organized by the Western States Section of the Combustion Institute and hosted by the University of Utah May 19-22, 213 Impact of Selective Oxygen Injection on NO, LOI, and Flame Luminosity in a Fine Particle, Swirl-Stabilized Wood Flame Joshua Thornock 1, Dale Tree 1, Yuan Xue 2, Vijaykant Sadasivuni 2, Remi Tsiava 3 1 Brigham Young University, Department of Mechanical Engineering, Provo, UT 8462, USA 2 Liquide, Delaware Research and Technology Center, Newark, DE 1972, USA 3 Liquide, Centre de Recherche Claude Delormel, Jouy en Josas, France Wood biomass is a renewable, carbon dioxide neutral fuel that has no sulfur and mercury emissions. Co-firing of wood biomass with coal allows the benefits of large scale high efficiency power plants to be obtained without the need for a continuous supply of fuel. One issue with wood as a fuel however, is that biomass particles are typically ground to a larger size than coal particles and can therefore produce poor burnout and elongated flames. Changes in flame shape and burnout alter the heat flux profile of a boiler making intermittent use of biomass more problematic. This paper investigates the use of strategic oxygen enrichment to reduce flame length and improve particle burnout. Measurements of exhaust NO, loss on ignition (LOI) and natural flame luminosity were obtained from a 15 kw, down-fired burner with oxygen enrichment. The fuel burned was a finely ground (224 m) hardwood. Flame images were obtained with a calibrated RGB digital camera allowing a calculation of visible radiative heat flux. The results showed that oxygen can be both beneficial and detrimental to the flame length depending on the momentum of the oxygen jet. Generally, the addition of oxygen increased emissions of NO but had little impact on LOI. Flame images suggest a small amount of oxygen can be used to stabilize the flame and increase devolitilization rates from the biomass with only minor increases to NO. 1. Introduction The co-firing of biomass and coal is an effective way to reduce net CO2 emissions and often decreases NOx and SOx as well. Most biomass is not available in particles sizes as small as that of coal because of the energy cost of the size reduction. The larger size of the biomass particle gives rise to a concern regarding longer residence times and decreased particle burnout [1]. Biomass char can also be less reactive than coal char. Thus, a blend of biomass and coal will increase the potential for extinction which may in turn lead to decreased particle burnout [2]. However, studies have shown increasing the oxygen concentration above that of air can have a positive effect on burnout in co-firing applications [3] [4]. This work uses selective oxygen injection to study the effects of higher oxygen concentrations on biomass burnout. While increasing oxygen concentration may increase particle burnout, the impact of increased O 2 on NO emissions and the radial heat flux profile is also important. The objective of this work is to measure the impact of O 2 addition to the burnout, NO and axial radial heat flux on a swirl-stabilized biomass flame. 2. Methods A 15 kwth, down-fired, Burner Flow Reactor (BFR) was used in conjunction with a burner designed and manufactured by Liquide. The BFR has a height of 2.4 m, an inside diameter of.75 m, and consists of six vertical,.4 m sections. These sections have four access ports 9 degrees apart. The burner was mounted at the top with exhaust exiting the
bottom. The Liquide burner is a pipe-in-pipe design allowing four separate inlet flows. Figure 1 shows the geometry of the four concentric pipes. Flow in channels 2 and 3 were held constant throughout the experiment with 2 kg/hr of air in channel 2 and 28 kg/hr air and 24 kg/hr of biomass in channel 3. Flow channel 4 contained secondary air which was adjusted in each case to keep the dry O 2 exhaust concentration at 4.%. Flows 3 and 4 have the option of swirl vanes that extend 4 inches into the pipe from the burner exit. The diameter of flow channel 1 was adjustable to allow varying oxygen injection velocities for the same oxygen flow rate. The burner configuration or burner geometry was characterized by a set of three variables: secondary swirl /primary swirl/flow channel 1 diameter. For example, SW/PS/dsmall represents a swirl vane on the secondary flow channel 4, swirl for primary air flow channel 3, and a small center tube diameter. Figure 1. Diagrams of the Liquide burner. (a) vertical cross-section showing the 4 flows exiting the burner A set of burner flow rates and burner configurations or geometries were selected as shown in Table 1. The test matrix involved changing the flow of gas in channel 1 from 2 kg/hr air to increasing amounts (2, 8, 18 kg/hr) of oxygen. This sweep of oxygen flow rates was studied for three different burner geometries. The first has swirl vanes located in both the secondary air (flow channel 4) and the primary air/fuel inlet (flow channel 3). The center tube was a relatively small diameter (). The second burner configuration consisted of swirl of the secondary air, no swirl of the primary and a small center tube (). The final burner configuration was to increase the diameter of the center tube while leaving only the secondary air swirled (). Table 1. Test matrix of operating conditions. Secondary swirl (SW), primary Swirl (PW), no swirl (NS), small diameter (d-small), and large diameter (d-large). Abbreviation Flow rates (kg/hr) (inlet 1) Burner Configuration 1 2(air); 2, 8, 18 ( ) Burner Configuration 2 2(air); 2, 8, 18 ( ) Burner Configuration 3 2(air); 2, 8, 18 ( ) Flame images were collected through the open reactor ports which are 92 mm wide by 29 mm in vertical length using a UNIQ, UC-6CL, 1-bit, RGB, CCD camera. The camera was calibrated for capturing absolute intensity measurements in red, green and blue pixels and used to obtain two dimensional maps of flame temperature and flame intensity. Draper et al. [5] and Svensson et al. [6] describe how to obtain temperature maps while absolute intensity values were estimated by rearranging an equation taken from Draper et al. [5] into the following equation: 2
(1) is the digitized pixel count of a color, is the aperture area, is the sensitivity constant for a given color i, is the spectral emissive power, is the spectral response of the optical system for a given color, and is the spectral transmittance of lenses and filters. In short, dividing the red pixel count by the aperture area, sensitivity constant for red, and the exposure time for an image provided an absolute intensity. NO measurements were collected from the exhaust line using a Horiba gas analyzer measuring (galvanic cell), (ND-IR), (ND-IR), and (chemiluminescence). Ash was also collected for each condition from a cyclone at the base of the BFR. Loss on ignition (LOI) defined by Eq. 2 were measured using the ASTM D7348 standard test method. In Eq. 2, is the weight of the moisture free ash and is the weight of the carbon free ash. (2) The fuel used was a finely ground hardwood with proximate and ultimate analysis as shown in Table 2. The particle size distribution was measured and produced a mean equivalent diameter of 224 m. This is a smaller diameter that is typically used in commercial biomass operation but larger than the typical size of coal particles. Table 2. Proximate and ultimate analysis of biomass (hardwood) used in all burner configurations Ultimate analysis, as received (wt.%) Proximate analysis, as received (wt.%) Moist 5.7 Moist (wt.%) 5.7 H 5.43 Ash (wt.%).63 C 49.17 Volatile (wt.%) 76.48 N.63 Carbon (wt.%) 17.82 S.4 O 39.3 HHV (kj/kg) 18,189 Ash.63 3. Results and Discussion Flame images were taken from the three upper access ports on the north side of the reactor. These ports vertically span distances of.3-.33 m,.45-.73 m, and.85-1.13 m from the burner. Typically a set of 4 images were taken at a given operating condition with a significant amount of variability from one image to the next caused by the turbulent nature of the flame. After viewing all of the images, a single image was selected that was visually representative of the average. This method is therefore subjective but is useful in providing a general description of the flame. More quantitative data will be shown in subsequent analysis. Flame images from the top port, spanning a distance from.3 m to.33 m below the burner exit are shown in Table 3. Each row of the table shows a constant set of flow rates and each column shows a constant geometrical configuration of the burner. The gain above each image compares the product of the aperture area and exposure time with the lowest value of this product which is set to 1.. A high gain indicates that an image is not as bright as it appears in comparison to a low gain image. The first row of images represents air only combustion while the second row has 2 kg/hr of oxygen added to the center burner tube. The flames in the first row appear to be brighter and more uniform than the flames in the second row unless the gain is considered. The image on the right end of the second row is shown twice in Table 4, once as shown in Table 3 and a second time with the intensity of all pixels increased. These images show that there are actually two flames zones in the 2 kg/hr images; a very intense flame surrounding the oxygen injection and a less intense flame similar to that of air-only combustion surrounding the oxygen flame. The introduction of oxygen in the center jet is therefore seen to produce a double flame. A central flame surrounding the incoming oxygen jet and an outer flame produced by mixing of the secondary air and fuel. 3
1 Table 3. Representative flame images for 4 oxygen center jet flow rates and 3 burner configurations GAIN=7.81 GAIN=6.25 GAIN=3.91 GAIN=1 GAIN=1 GAIN=1 GAIN=6.25 GAIN=1 GAIN=1 GAIN=7.81 GAIN=1 GAIN=1 4
Table 4. Comparison of the same image at, burner configuration. The image on the right has the intensity of all pixels increased showing the existence of a high intensity oxygen flame within the lower intensity air flame. Initial Image Brightness Enhanced Image The influence of the burner geometries on the center flame can be seen by comparing the inner flame size for 2 kg/hr of O 2 flow. The two flames on the left have the oxygen flowing through a smaller diameter tube than the flame on the right. This smaller diameter produces a higher velocity and elongates the inner oxygen flame in comparison to the flame imaged on the right. Moving down the table to higher flow rates of oxygen in the center tube, the ability to visually identify the inner oxygen flame from the air flame was more difficult. The inner flame was larger and penetrating the entire length of the top port due to the higher momentum in the oxygen jet but the boundary between the air and oxygen flame was not as sharp. This higher momentum appears to increase mixing between the oxygen and the surrounding gas. For the geometries represented in the two columns to the right, the ignition of the center flame is seen to be delayed with increasing O 2 flow rate allowing entrainment of fuel and other gases into the center O 2 jet prior to combustion. A final note from the flame images is that the geometry with swirl in both the fuel stream and the secondary air stream ( left column) produces flames of lower intensity (higher gain) than the geometries where the fuel stream was not swirled. This is consistent with increased mixing between the fuel and air streams which would reduce soot formation and flame temperature. Two techniques were used to produce quantitative data from the images. In the first, the calibrated camera was used to turn the digital pixel counts into a relative intensity as discussed in the method section. The objective was to produce an axial flame intensity from the burner to the end of the flame. This was accomplished by dividing an image into nine sections, each 3.5 cm in height. The pixels within each section of an image were averaged to create a single intensity representing that axial position. Each averaged intensity was then ensemble averaged with intensities from all of the images obtained which was typically between 3 and 75 images. The resulting intensity profiles obtained from images in the top three ports are shown in Figure 2 for each of the three burner geometries. The intensity plots shown in Figure 2 are consistent with the previous explanation of the flame images in Table 3. The intensity of the air-only flames is consistently lower than any of the oxy-fired cases. In each of the burner configurations the intensity of the case produces a very large peak indicative of the center oxygen flame seen in the images. This high intensity is seen for all three burner geometries and it is interesting to note that regardless of the high initial value the intensity quickly returns to an intensity level similar to the air case. As the flow rate of oxygen increases the location of the peak intensity moves downstream. The 8 and 1 cases reach a maximum toward the end of the first port and beginning of the second port respectively. The 18 kg/hr case 5
Intensity (W/m^2) Intensity (W/m^2) Intensity (W/m^2) reaches the highest intensities of all flow rates and produces higher intensities downstream in the third port. The two higher oxygen flow rates appear to ignite later and produce lower intensities than 2 kg/hr in the vicinity near the burner. 1 9 8 7 6 5 4 3 2 1 1 9 8 7 6 5 4 3 2 1 1 9 8 7 6 5 4 3 2 1 (a) (b) 1 1 1 (c) Sw/NS/d-large Figure 2. Intensity [W/m^2] vs. distance [cm] from the burner of three burner configurations: (a) SW/PS/dsmall, (b), and (c) SW/PS/d-large. 6
Intensity (W/m^2) Intensity (W/m^2) Intensity (W/m^2) Intensity (W/m^2) The same intensities shown in Figure 2 are rearranged in Figure 3 to more clearly show the differences between burner configurations. Independent of oxygen flow rate, the configuration with the most mixing () had the lowest intensity while the configuration with least mixing (SW/PS/d-large) had the highest intensity. For each flow condition the configuration has a more rapidly declining intensity after the peak. 3 25 2 15 1 5 (a) 1 9 8 7 6 5 4 3 2 1 (b) 2 kg/hr center oxygen 1 9 8 7 6 5 4 3 2 1 (c) 8 kg/hr center oxygen 1 9 8 7 6 5 4 3 2 1 (d) 18 kg/hr center oxygen Figure 3. For each air flow configuration, the axial intensity is shown for each burner configuration. The second method used to analyze the flame images was to calculate the temperature. Figure 4 shows the temperature data of the three burner configurations. As might be expected, the air-only flow configuration produced the lowest flame temperature. Although not always true, the trend in the top port between.3 and.33 m, is that the temperature increases with increasing oxygen flow rate. The oxygen flame temperatures are highest in the burner configuration with primary swirl (Figure 4a) in comparison to the other two configurations (Figure 4b and 4c). This was the same burner configuration shown in Figure 2 with lower intensity indicating less soot and perhaps less heat loss from the flame due to radiative heat transfer. The lower heat loss from the flame would explain the higher temperature. The higher swirl burner configuration (Figure 4a) also produced the largest temperature difference between the oxygen and air-only cases. Also in Figure 4a it can be seen that the flow condition begins near the burner at a higher temperature than the air flow condition but decreases temperature more rapidly with increasing distance from the burner. For some flow conditions, the temperature could not be determined because the intensity was too low to be measured or to produce a measurable temperature. The data in Figure 4 support the conclusion that burner configurations with less overall mixing result in lower near burner temperatures and higher downstream temperatures for the oxygen injected cases. It is also interesting to note that the intensity peak does not correlate to a higher temperature. 7
Temperature (K) Temperature (K) Temperature (K) 23 22 21 2 19 18 17 16 15 14 13 23 22 21 2 19 18 17 16 15 14 13 23 22 21 2 19 18 17 16 15 14 13 Distance from the burner (cm) (a) (b) 1 1 Distance from the burner (cm) 1 Distance from the burner (cm) (c) Figure 4. Temperature [K] vs. distance [cm] from the burner exit of three burner configurations: (a) SW/PS/dsmall, (b), and (c). 8
LOI (%) NO (ppm) The flue gas concentrations of NO are shown for each of the flow conditions and burner configurations in Figure 5. Error bars on the figure show plus or minus one standard deviation where more than one replication of the operating condition were obtained. The results show a general trend of increasing NO concentration with increasing oxygen flow rate where zero oxygen flow rate represents the air-only flow condition. A small amount of oxygen injection (2 kg/hr) shows little if any increase in NO, while further increases in NO produce more significant increases in NO. Given the flame images shown in Table 3, the 2kg/hr flow rate of oxygen produced a small flame surrounding the oxygen injection which was not well mixed with the fuel. The high temperature produced by this inner flame may increase the devolatilization rate of the fuel without significantly increasing oxygen availability to the fuel and therefore, NO formation is not increased. With higher O2 flow rates, the mixing between fuel and oxygen prior to ignition appears to increase which is likely the cause of the increased NO concentrations. Loss on ignition data for each of the flow conditions and each burner configuration are shown in Figure 6. The loss on ignition is seen to be very low for this biomass, probably do to the fine particle size of the fuel. There does not appear to be a strong trend with LOI and oxygen flow rate, but instead, the LOI correlated more closely with the burner configuration. The lowest LOI was obtained with the burner configuration that had the highest swirl and showed indications of being the most well mixed. This is consistent with the fact that char burnout is typically diffusion limited, not temperature limited. Therefore higher concentrations of surrounding oxygen should produce better burnout. The difference in LOI from the other burner configurations is not well understood except to say that differences in all of the LOI are small and burnout was very good. 3 25 2 15 1 5 5 1 15 2 Oxygen Flow (kg/hr) Figure 5. Nitrogen oxide concentrations as a function of oxygen flow rate in the center tube. 18 16 14 12 1 8 6 4 2 5 1 15 2 Oxygen Flow (kg/hr) Figure. 6. Loss on ignition (LOI) as a function of oxygen flow for each of the burner configurations. 9
4. Summary and Conclusions A burner capable of various oxygen injection strategies was used to study the impact of oxygen injection flow rate and momentum on a biomass flame with an average particles size 224 microns. The flame shape, intensity and temperature were examined using a two-color digital camera. Flame images, NO, and LOI data were obtained for a matrix of 12 operating conditions of variable oxygen flow rate and burner geometries. The addition of a small flow rate of oxygen (2 kg/hr) was found to produce a second oxygen rich flame in the center of an air flame. This oxygen flame was of very high intensity and temperature (21 K). The presence of this flame did not increase NO or change burnout but increased the near burner heat flux. Increasing the oxygen flow rate to 8 and 18 kg/hr produced an elongated oxy-flame that produced higher temperatures and intensity over the length of the luminous signal. The NO was found to increase with increasing oxygen flow rate but burnout was not improved, possibly due to the high burnout rate already achieved with air combustion. Swirling the fuel stream in addition to the secondary air stream produced increased mixing, lower intensity and higher temperatures due to a reduced amount of soot formation. The addition of oxygen increased flame temperatures near the burner but the temperatures dropped more rapidly than the air-only flames. Acknowledgements This research was funded by Liquide USA. References [1] Larry Baxter, Green Energy and Technology 211, pp 43-73. [2] Behdad Moghtaderi, Fuel 86 (27) 2431-2438 [3] B. Arias, C. Pevida, F. Rubiera, J.J. Pis, Fuel 87(28) 2753-2759 [4] John P. Smart, Rajeshriben Patel, Gerry S. Riley, Combustion and Flame 157 (21) 223-224 [5] T.S. Draper, D. Zeltner, D.R. Tree, Y. Xue, R. Tsiava, Appl. Energy 95 (212) 38-44. [6] K.I. Svensson, A.J. Mackrory, M.J. Richards, D.R. Tree, SAE Paper 25-1-648,25. 1